These lemmas help determining if a representation is irreducible or not.
Schur's lemma I
group, finite dimensional vector space and an irreducible representation.
If linear operator is such that then (i.e. a multiple of the identity).
Short: if a linear operator commutates with an irreducible representation, it can only be a multiple of the identity.
Conclusion
If we are unsure if is irreducible or not and we find an (not a multiple of the identity) that commutes with all the 's, then must be reducible.
Consequences
Every Abelian (commutative) group has only 1D irreducible representations
The fact that can be diagonalized globally is not surprising anymore
Every irreducible representation has since it commutes with anything
Take a look at : is it also an eigenvector of ? because of commutativeness, so is also an eigenvector or with the same eigenvalue it is in the same eigensubspace of as .
Then the eigensubspace of is an invariant subspace of the representation as well2. But is an irreducible representation, so it can only have one invariant subspace, that is the whole space . It can only commute with the identity matrix or a multiple of it.
Schur's lemma II
group, and vector spaces and irreducible representations.
If linear operator is such that then is either or invertible (then ).
Short: if and , then .
Proof
Let's take a look at and .
is an invariant subspace of and since is irreducible, or .
.
Then is an invariant subspace of and since is irreducible, or .
Commutative matrices have common invariant subspaces because two matrices that are simultaneously diagonalizable are always commutative. See Commuting matrices ↩